Can I feed bypass clock source to HSE clock input in ceramic/crystal resonator mode ?

[Long Pham]

Posted on July 06, 2018 at 17:35

I'm making a proto board but I don't have suitable loading capacitance for crystal resonator.

As far as my novice knowledge about Pierce oscillator and the fact that some other MCUs may not care which clock source fed to them, I think it's possible.

Is it possible to do so?

#hse-config #external-clock-source #hse

[Tesla DeLorean]

Posted on July 06, 2018 at 18:28

If you use an external clock source (XO, TCXO, etc) you should generally try to use HSE BYPASS mode. There are some STM32 family that will tolerate you not doing that, but I've observed the L1 parts to have a halving/doubling issue, and the L0 where BYPASS doesn't work at 32 MHz

[Long Pham]

Posted on July 07, 2018 at 11:34

I should have added this information. I'm going to test DFU bootloader (STM32F4) . In the 'bootloader selection' block diagram (AN2606), it seems that it need 'HSE detected' to execute DFU bootloader. I don't know if this includes bypass clock source or not.

[LMI2]

Posted on July 07, 2018 at 15:01

''m making a proto board but I don't have suitable loading capacitance for crystal resonator.' Those capacitors are for fine tuning the crystal frequency. They may also an effect of loading internal electronics of the CPU. There is always some capacitance between IC pin and ground. So in that sense those capacitors are not critical or even needed. But as Clive2version0 said there maybe other issues.

[Tesla DeLorean]

Posted on July 07, 2018 at 17:30

I don't have a good way to test it, but my expectation is that HSE_READY is driven by the presence of a clock on OSC/HSE_IN pin, and then it bench marks that clock against the HSI via a TIM, to determine which of the specifically supported clock frequencies is being used.

[Long Pham]

Posted on July 08, 2018 at 16:32

According to this

https://electronics.stackexchange.com/questions/94031/what-effect-does-selecting-the-wrong-load-capacitance-for-a-crystal-have

, it might be bad if I just rely on parasitic capacitance.

[Long Pham]

Posted on July 08, 2018 at 16:40

Looking at the schematic of STM32F746G-DISCO board, I can see that ST uses a crystal oscillator as the clock source for the chip. Well, that's the motivation to test my F4 board.

[LMI2]

Posted on July 08, 2018 at 17:47

If this is your first board and you have no measuring equipment, yes do everything by the book. Especially if you are not going to make several prototypes to test everything properly.

[Long Pham]

Posted on July 08, 2018 at 18:10

I have an old PIC board having no loading capacitance but it worked (lol).

[LMI2]

Posted on July 08, 2018 at 22:38

Yep. I have done the same with some Atmel CPUs without problems. But Clive2 has more experience from STM32 CPUs than me.